Inconsistencies with different Yambo Versions

Deals with issues related to computation of optical spectra in reciprocal space: RPA, TDDFT, local field effects.

Moderators: Davide Sangalli, andrea.ferretti, myrta gruning, andrea marini, Daniele Varsano, Conor Hogan

Post Reply
batman
Posts: 12
Joined: Sun Jun 20, 2021 2:04 pm

Inconsistencies with different Yambo Versions

Post by batman » Wed May 31, 2023 4:09 pm

Dear Yambo Developers,

The BSE results from different Yambo version 5 releases are not consistent with each other even when the input files are exactly the same. I attached an image for Silicon done with different versions.

i) Firstly, for Version 5.1.2, the results with and without terminator do not match (both are converged w.r.t to bands) but when done with 5.0.4 using same input files, both with and without XTermKind='BG' gave the same results (as shown in the plot).

ii) Last and mostly importantly, the BSE spectrum does not agree with different versions (also shown in the figure). Both yambo versions gave me different results.

Could you please comment on why do I see these discrepancies and what version to stick to?. I attached the report and input files (input files are same for all cases except changing the XTermKind parameter).

The Job names for attached files:
withBG. -> 5.1.2 with XTermKind = 'BG'
withnoBG -> 5.1.2 with out XTermKind = 'BG'
withBG504 -> 5.0.4 with XTermKind = 'BG'
withnoBG504 -> 5.0.4 with out XTermKind = 'BG'

Thank you,

Best regards,
Murali
You do not have the required permissions to view the files attached to this post.
Last edited by batman on Wed May 31, 2023 4:20 pm, edited 3 times in total.
Muralidhar Nalabothula
Doctoral student at Department of Physics and Materials Science,
Université du Luxembourg

batman
Posts: 12
Joined: Sun Jun 20, 2021 2:04 pm

Re: Inconsistencies with different Yambo Versions

Post by batman » Wed May 31, 2023 4:16 pm

Dear Daniele,

updated the post.
Muralidhar Nalabothula
Doctoral student at Department of Physics and Materials Science,
Université du Luxembourg

User avatar
Daniele Varsano
Posts: 3773
Joined: Tue Mar 17, 2009 2:23 pm
Contact:

Re: Inconsistencies with different Yambo Versions

Post by Daniele Varsano » Wed May 31, 2023 4:19 pm

Thank you very much,
we will have a look soon.
Please notice, that the Xterm usage is discouraged and probably it will be dismissed, as it has a large pre-factor and finally it is not efficient, contrary to the Gterm.

Daniele
Dr. Daniele Varsano
S3-CNR Institute of Nanoscience and MaX Center, Italy
MaX - Materials design at the Exascale
http://www.nano.cnr.it
http://www.max-centre.eu/

User avatar
Davide Sangalli
Posts: 610
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: Inconsistencies with different Yambo Versions

Post by Davide Sangalli » Thu Jun 01, 2023 8:04 am

Dear Murali,
I tested this in serial, and 5.0, 5.1 and 5.2 always give the same result.

I wonder if the issue is due to the parallelization.
Can you try to reduce the parameters of your simulation and run in serial to check if also in such case you experience the same issue?

Best,
D.

P.S.: In case you saw the initial post asking for logs and report of 5.0, I figured out that they are in the zip of 5.1
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

batman
Posts: 12
Joined: Sun Jun 20, 2021 2:04 pm

Re: Inconsistencies with different Yambo Versions

Post by batman » Thu Jun 01, 2023 9:56 am

Dear Davide,

Thanks for the reply. I ran on single core and they do not match. I am posting the report files and inputs that I ran on single core. The results on single core are consistent with the multicore ones. I used exactly same input file in both cases (without terminator in both cases)



Best regards,
Murali
You do not have the required permissions to view the files attached to this post.
Muralidhar Nalabothula
Doctoral student at Department of Physics and Materials Science,
Université du Luxembourg

User avatar
Davide Sangalli
Posts: 610
Joined: Tue May 29, 2012 4:49 pm
Location: Via Salaria Km 29.3, CP 10, 00016, Monterotondo Stazione, Italy
Contact:

Re: Inconsistencies with different Yambo Versions

Post by Davide Sangalli » Thu Jun 01, 2023 10:15 am

Ok, then I've really no idea of what is happening.
Also inspecting the two reports, all parameters are the same.

Checks which you could do:
a) inspect the ndb.em1s* databases with ncdump and see if they are different or the same
b) run 5.0 on the screening computed with 5.1 and viceversa (you might have issues with the header)
c) try a different BSE solver

Best,
D.
Davide Sangalli, PhD
CNR-ISM, Division of Ultrafast Processes in Materials (FLASHit) and MaX Centre
https://sites.google.com/view/davidesangalli
http://www.max-centre.eu/

batman
Posts: 12
Joined: Sun Jun 20, 2021 2:04 pm

Re: Inconsistencies with different Yambo Versions

Post by batman » Thu Jun 01, 2023 1:51 pm

Dear Davide,

a & b) I tried to swap the screening (use screening from 5.1.2 and do bse with 5.0.4), Now they match!. So I think, the screening is fine.
Also they only match when I do not use X-terminator. So I guess the X terminator is broken in the newer version. Please note that I have to recompute the dipoles again as it complains about database mismatch.

c) I tried different solver, both are not same. So I guess it is not an issue with the solver (as shown in the plot) .

Best regards,
Murali
You do not have the required permissions to view the files attached to this post.
Muralidhar Nalabothula
Doctoral student at Department of Physics and Materials Science,
Université du Luxembourg

Post Reply